more gcj 3.0 fun -- catch { try {...}...}, encodings

[Alexandre Petit-Bianco]

David Brownell writes:

> Oooh, I want it!  Didn't see it on java-patches (archive) yet
> though.  That was indeed the same bug, I take it?

I'm still testing it. It's in GNATS.

> All that "xml.vendor" stuff is gone from the ClasspathX version.
> No need for it any more.

Good.

./A

2001-07-11  Alexandre Petit-Bianco  <apbianco@redhat.com>

	* parse.y (verify_constructor_super): Compare anonymous class ctor
	args with `valid_method_invocation_conversion_p.'

Index: parse.y
===================================================================
RCS file: /cvs/gcc/gcc/gcc/java/parse.y,v
retrieving revision 1.294
diff -u -p -r1.294 parse.y
--- parse.y	2001/07/11 07:33:43	1.294
+++ parse.y	2001/07/12 05:56:02
@@ -8656,7 +8656,9 @@ verify_constructor_super (mdecl)
 		  && m_arg_type != end_params_node);
 		 arg_type = TREE_CHAIN (arg_type), 
 		   m_arg_type = TREE_CHAIN (m_arg_type))
-	      if (TREE_VALUE (arg_type) != TREE_VALUE (m_arg_type))
+	      if (!valid_method_invocation_conversion_p 
+		     (TREE_VALUE (arg_type),
+		      TREE_VALUE (m_arg_type)))
 		break;
 
 	    if (arg_type == end_params_node && m_arg_type == end_params_node)